Bank examines SpaceX plan for 80,000 cell-towers

A report from investment bank BNP Paribas examines in detail the Bloomberg report that SpaceX is in talks with US cell-tower operators and is seeking a minimum of 80,000 cell towers. The bank has looked at the cost – and likelihood – of the plan, and whether this confirms SpaceX’s ambition to launch a full cellular service to the US.

The bank’s report, from senior analyst Sam McHugh, is blunt, saying: “It’s impossible to know if SpaceX is still bluffing for an MVNO (Mobile Virtual Network Operator) here, but it is the last stop on that train. Once SpaceX signs “take or pay” Master Lease Agreements with the tower companies committing to a certain minimum number of site deployments, they can’t cancel these contracts without payments on the scale of the lifetime lease costs. If it is going to go down the path of building a “capitally efficient” network by relying on an MVNO, it must make that choice now (and certainly by the time the upper C-Band auction comes around in April 2027).”

“The stakes are underappreciated,” suggests BNP Paribas. “Before SpaceX could capture a dollar of revenue for its mobile network, without an MVNO it would need to spend at least $100 billion. This excludes the cost of any additional spectrum (which, in this scenario, they would need), as well as the cost of the capital needed to fund $100 billion, before even considering the political/regulatory landscape could change a lot over the next few years. In engaging with the tower companies, it is also increasingly clear SpaceX does not have a huge network cost advantage from their satellites, if anything there is somewhat of a disadvantage of having to operate both terrestrial and Space infrastructure.”

“SpaceX has been banking on an MVNO. Since the beginning of the year investor feedback has consistently been that SpaceX expect to get an MVNO in the US, and their strategy at the beginning looked a lot like the route the Cable companies used (buy Spectrum and use as a bargaining chip). But the telcos are not Spectrum constrained like Verizon was back in the early 2010’s, and so this strategy has not worked, leaving SpaceX with the unenviable task of building bottom up if they have a genuine ambition of being a 4th carrier,” the ban says.

“The Telcos shouldn’t budge, but this does create more headline risk,” advises the bank. “We continue to think the telcos are better off not folding on an MVNO deal (it would only enable SpaceX to then build more economically), but this does come with heightened risk that SpaceX try to exert political pressure post the mid-terms (discussed in our note last night, re potential arguments around a “concerted refusal to deal” or attacking the Satellite JV construct – although the outcome of any potential litigation could take years, and still not result in any kind of MVNO/roaming remedy).”

The bank suggests that this could end with a different kind of deal with SpaceX? “Listening to SpaceX’s 3Q26 earnings call, it seems clear that a big part of their ambition on building a network is to connect a swathe of Physical AI devices in the long term (cars, robots etc), and there’s still a case where what they build doesn’t serve consumers, but is instead focused on this opportunity. If we were a carrier, and we think TMUS is uniquely placed here with their more advanced 5G network and relatively low broadband exposure, we would be tempted to offer SpaceX a deal – agree to wholesale a “dedicated network slice” for future AI needs (excluding phones), and to collaborate on development here, while also agreeing to resell Starlink broadband alongside T-Mobile wireless in stores and online – offering SpaceX solutions to both the AI network needs and also their convergence disadvantage.”

BNP Paribas adds that achieving maximum distribution terrestrially would mean that the 80,000 tower sites would be the first step. It reminds investors that AT&T and Verzon have about 65,000-70,000 towers in use, while T-Mobile has some 85,000 towers. The banks says: “If SpaceX stood up a network with 80k macro sites, its coverage would be on par with the Big Three. It is a dense enough grid to carry the C-Band. A buildout to 80k towers that then seamlessly combines with an advanced network of 15,000 mobile satellites to create continuous, ubiquitous coverage has a shot at being differentiated enough to make it – albeit without the Spectrum depth of the carriers, which would ultimately limit SpaceX’s subscriber share potential.”

But the bank also cautions that if the Bloomberg report is accurate, and 80,000 towers are the ambition then it would have to be more than one supplier/partner company and cites American Tower and Crown Castle as probably being essential.

Categories: Blogs, Business, Inside Satellite, Satellite

Creative Media Centre
Hastings
East Sussex
TN34 1HL

Tel: +44 7900 196870